US20140049281A1 - Diagnosis Framework to Shorten Yield Learning Cycles of Advanced Processes - Google Patents
Diagnosis Framework to Shorten Yield Learning Cycles of Advanced Processes Download PDFInfo
- Publication number
- US20140049281A1 US20140049281A1 US13/588,155 US201213588155A US2014049281A1 US 20140049281 A1 US20140049281 A1 US 20140049281A1 US 201213588155 A US201213588155 A US 201213588155A US 2014049281 A1 US2014049281 A1 US 2014049281A1
- Authority
- US
- United States
- Prior art keywords
- dut
- test
- output
- yield
- carrier unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims description 31
- 230000008569 process Effects 0.000 title claims description 15
- 238000003745 diagnosis Methods 0.000 title abstract description 11
- 238000012360 testing method Methods 0.000 claims abstract description 168
- 230000007547 defect Effects 0.000 claims abstract description 90
- 238000004458 analytical method Methods 0.000 claims abstract description 51
- 239000013598 vector Substances 0.000 claims description 76
- 238000001465 metallisation Methods 0.000 claims description 32
- 238000005259 measurement Methods 0.000 claims description 13
- 238000004088 simulation Methods 0.000 claims description 11
- 238000013461 design Methods 0.000 claims description 10
- 239000000463 material Substances 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 abstract description 17
- 238000005516 engineering process Methods 0.000 abstract description 15
- 230000015572 biosynthetic process Effects 0.000 abstract description 4
- WYTGDNHDOZPMIW-RCBQFDQVSA-N alstonine Natural products C1=CC2=C3C=CC=CC3=NC2=C2N1C[C@H]1[C@H](C)OC=C(C(=O)OC)[C@H]1C2 WYTGDNHDOZPMIW-RCBQFDQVSA-N 0.000 description 10
- 239000011159 matrix material Substances 0.000 description 5
- 230000007704 transition Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 230000001066 destructive effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012797 qualification Methods 0.000 description 2
- 238000013102 re-test Methods 0.000 description 2
- 238000012356 Product development Methods 0.000 description 1
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 1
- 230000003466 anti-cipated effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000001010 compromised effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 230000003278 mimic effect Effects 0.000 description 1
- 229910052710 silicon Inorganic materials 0.000 description 1
- 239000010703 silicon Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/28—Testing of electronic circuits, e.g. by signal tracer
- G01R31/317—Testing of digital circuits
- G01R31/31704—Design for test; Design verification
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/28—Testing of electronic circuits, e.g. by signal tracer
- G01R31/317—Testing of digital circuits
- G01R31/3181—Functional testing
- G01R31/3183—Generation of test inputs, e.g. test vectors, patterns or sequences
- G01R31/318342—Generation of test inputs, e.g. test vectors, patterns or sequences by preliminary fault modelling, e.g. analysis, simulation
- G01R31/318357—Simulation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/28—Testing of electronic circuits, e.g. by signal tracer
- G01R31/2851—Testing of integrated circuits [IC]
- G01R31/2884—Testing of integrated circuits [IC] using dedicated test connectors, test elements or test circuits on the IC under test
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/52—Testing for short-circuits, leakage current or ground faults
Definitions
- FIG. 1 illustrates a flow diagram of some embodiments of a method to locate and diagnose defects in an integrated circuit.
- FIG. 2 illustrates some embodiments of a tool arrangement configured to locate and diagnose defects in an integrated circuit.
- FIG. 3 illustrates some embodiments of a yield learning circuit structure.
- FIG. 4 illustrates some embodiments of a yield learning circuit structure.
- FIG. 5 illustrates some embodiments of a DUT carrier circuit.
- FIG. 6A illustrates some embodiments of a DC yield learning circuit structure.
- FIG. 6B illustrates some embodiments of a look-up table comprising expected vs. simulated results
- FIG. 7A illustrates some embodiments of an AC yield learning circuit structure.
- FIG. 7B illustrates some embodiments of an AC look-up table.
- FIG. 8A illustrates some embodiments of a discrete library of allowable patterns for technology qualification comprising a process sensitive component.
- FIG. 8B illustrates some embodiments of a 4-to-1 XOR tree comprising a discrete library of allowable patterns.
- Diagnosis of defects in early technology node test vehicles allows for improved yield by determining a failure mode (i.e., root cause) associated with a manufacturing defect, which may be addressed through a change in design, a change in a manufacturing process, or a combination thereof.
- Current diagnosis methods utilize test patterns to observe defects, and allow for one to draw inferences based on observed behavior which can narrow the search for a failure mode.
- PFA physical failure analysis
- PFA physical failure analysis
- several iterations of test pattern application, result interpretation, and making inferences may be necessary to determine a region where PFA may be utilized to determine an exact location and failure mode of a defect. This can limit the overall efficiency in diagnosing all failure modes, which can increase the lead time for product development and mask reliability issues which may only be discovered once a product is in the field. As such, increased efficiency in defect diagnosis is desired.
- a diagnosis framework to shorten yield learning cycles of technology node manufacturing processes from a high defect density stage to technology maturity.
- a plurality of defect under test (DUT) structures are designed to capture potential manufacturing issues associated with defect formation.
- a test structure is formed by arranging the DUT structures within a defect under test (DUT) carrier unit, which has been yield-hardened though heuristic yield analysis such that a defect density of the DUT carrier unit is essentially zero.
- Possible outcomes of an application of test patterns and various failure scenarios associated with defects formed within the DUT structures within the DUT carrier unit are simulated and stored in a look-up table (LUT).
- LUT look-up table
- the LUT may then be referenced to determine a location of a defect within the test structure without the need for iterative analysis to correctly select defect candidates for physical failure analysis (PFA).
- PFA physical failure analysis
- FIG. 1 illustrates a flow diagram of some embodiments of a method 100 to locate and diagnose defects within an integrated circuit. While method 100 is illustrated and described below as a series of acts or events, it will be appreciated that the illustrated ordering of such acts or events are not to be interpreted in a limiting sense. For example, some acts may occur in different orders and/or concurrently with other acts or events apart from those illustrated and/or described herein. In addition, not all illustrated acts may be required to implement one or more aspects or embodiments of the description herein. Further, one or more of the acts depicted herein may be carried out in one or more separate acts and/or phases.
- a reusable defect under test (DUT) carrier unit comprising a plurality of logic devices.
- the logic devices within the DUT carrier unit have been subjected to heuristic yield analysis comprising manufacture and test of the logic devices to identify defects formed within a respective logic device, physical failure analysis (PFA) to locate the defects and assign a failure mode to a respective defect, and have been yield-hardened through re-design, re-manufacture, and re-test.
- the heuristic yield analysis may comprise several iterations of re-design, re-manufacture, and re-test to insure robust yield.
- the DUT carrier unit may be generically designed and reused for an arbitrary collection of DUTs.
- a respective DUT structure comprises at least one process sensitive component formed on one or more of a front end of line gate material layer, a middle end of line interconnect layer, or a back end of line metallization layer.
- a netlist comprising a definition of connections between the logic devices within the test structure formed by the plurality of DUT structures is compiled.
- the netlist may be compiled using an industry-standard extraction tool or an in-house netlist compiler.
- a respective test vector comprises a multi-bit binary test vector which serves as an input to the test structure, and may be generated through utilization of an automatic test pattern generation (ATPG) tool to mimic a manufacturing defect within the test structure.
- AGT automatic test pattern generation
- expected outputs of applying the plurality of test vectors to the test structure are simulated to define possible failure locations within the test structure.
- the simulations may be performed using an industry-standard a fault simulation tool or design for test (DFT) tool, and are stored in a look-up table (LUT).
- DFT design for test
- test structure is manufactured, which comprises fabrication of a silicon-based early technology yield learning vehicle in the 20 nm technology node, the 14 nm technology node, or lower technology node.
- the plurality test vectors are applied to the test structure to generate measured outputs.
- Application of the plurality test vectors comprises multi-bit binary test vectors scanned-in through a first shift register, operated on by the logic devices within the test structure to generate the measured outputs, which are then scanned out through a second shift register.
- the measured outputs are compared to the expected outputs to determine a failure location within the test structure, which comprises locating the stuck bits within the LUT.
- PFA physical failure analysis
- the failure mode may be cataloged for future reference. Note that PFA is not necessary to determine the location of the fail as that is determined by the look-up table. Moreover, PFA need not be performed if the failure mode has already been identified and cataloged.
- FIG. 2 illustrates some embodiments of a tool arrangement 200 configured to locate and diagnose defects in an integrated circuit.
- a test vector generator 204 is configured to define a test vector 206 corresponding to a test structure 202 .
- a fault simulation tool 208 is configured to receive the test vector 206 from the test vector generator 204 and to simulate an expected output 210 of applying the test vector 206 to the test structure 202 which corresponds to a failure location within the test structure 202 .
- a look-up table (LUT) 212 is configured to receive the expected output 210 from the fault simulation tool 208 and to store the expected output 210 for reference.
- LUT look-up table
- a measurement and analysis tool 214 is configured to receive a measured output 216 of applying the test vector 206 to the test structure 202 , and to compare the measured output 216 to the expected output 210 to determine a failure location within the test structure 202 without performing physical failure analysis (PFA).
- PFA physical failure analysis
- the test structure 202 comprises a first shift register 222 which is controlled by a clock 218 and configured to receive the test vector 206 from the test vector generator 204 .
- a controller 220 is configured to control operation of the test vector generator 204 and the clock 218 .
- the first shift register 222 comprises a plurality of scan registers 226 which are sequentially coupled and configured to receive the test vector 206 comprising a multi-bit binary vector as a serial data string and distribute the test vector 206 in parallel format to a DUT carrier unit 224 .
- a second shift register 228 which is controlled by the clock 218 is configured to receive a measured output 216 from the DUT carrier unit, and to send the measured output 216 to the measurement and analysis tool 214 .
- the second shift register 228 comprises a plurality of scan registers 226 which are sequentially coupled and configured to receive the measured output 216 from the DUT carrier unit 224 as a single bit, and to distribute the measured output 216 to a measurement and analysis tool 214 .
- the second shift register 228 may also receive multiple measured outputs 216 and distribute the multiple measured outputs 216 as a serial data string to a measurement and analysis tool 214 .
- the DUT carrier unit 224 comprises a plurality of logic devices which are coupled by a plurality of DUT structures to form a logic tree for which a known input (e.g., the test vector 206 ) will produce an expected output 210 .
- a known input e.g., the test vector 206
- the DUT carrier unit 224 comprises a plurality of logic devices which are coupled by a plurality of DUT structures to form a logic tree for which a known input (e.g., the test vector 206 ) will produce an expected output 210 .
- a known input e.g., the test vector 206
- a design tool 230 is configured to produce a design of the test structure 202 , and is coupled to a netlist compiler 232 configured to receive information about the design of the test structure 202 and to compile a netlist of the connections formed between the plurality of logic devices.
- the netlist compiler 232 is further configured to send the netlist to the fault simulation tool 208 in order to simulate the expected output 210 .
- the measurement and analysis tool 214 of the tool arrangement 200 can determine the location without PFA by utilizing the look-up table 212 to compare the expected output 210 to the measured output 216 .
- FIG. 3 illustrates some embodiments of a yield learning circuit structure 300 , comprising a first shift register 301 a, a second shift register 301 b, and a plurality of defect under test (DUT) carrier units 304 a - 304 j, wherein a respective DUT carrier unit of the plurality of DUT carrier units 304 a - 304 j comprises a plurality of logic devices arranged in a logic tree structure which has been yield-hardened.
- DUT defect under test
- the first shift register 301 a comprises a plurality of first scan registers 302 a - 302 o which are sequentially coupled, wherein a respective first scan register of the plurality of first scan registers 302 a - 302 o is connected to N DUT carrier units of the plurality of DUT carrier units 304 a - 304 j.
- Each output of a respective DUT carrier unit of the plurality of DUT carrier units 304 a - 304 j is connected to a respective second scan register of a plurality of second scan registers 314 a - 314 o of the second shift register 301 b, which selects an input source according to mode selection signal 310 .
- Each second scan register 314 a - 314 o comprises two data inputs.
- a first input of a respective second scan register of the plurality of second scan registers 314 a - 314 o is connected to an output of a respective DUT carrier unit of the plurality of DUT carrier units 304 a - 304 j.
- a second input of a respective second scan register of the plurality of second scan registers 314 a - 314 o is connected to an output of a previous second scan register of the plurality of second scan registers 314 a - 314 o (i.e., a second input of scan register 314 o is connected to an output of scan register 314 n, etc.).
- the first shift register 301 a and the second shift register 301 b each comprise two operation modes: a shift mode and a capture mode, which are controlled by a mode selection signal 310 .
- all second scan registers of the plurality of second scan registers 314 a - 314 o form a shift register between a primary input and a primary output
- each second scan register of the plurality of second scan registers 314 a - 314 o will receive an output value of a preceding second scan register of the plurality of second scan registers 314 a - 314 o.
- each second scan register of the plurality of second scan registers 314 a - 314 o will change to receive the output value of a respective DUT carrier unit of the plurality of DUT carrier units 304 a - 304 j.
- the first shift register 301 a will require M first scan registers of the plurality of first scan registers 302 a - 302 o
- the second shift register 301 b will each require M second scan registers of the plurality of second scan registers 314 a - 314 o.
- output data 312 of the first 8 first scan registers 302 a - 302 j are connected to 8 inputs of the first DUT carrier unit 304 a
- outputs of the 2nd through 9th first scan registers 302 b - 302 k are connected to the 8 inputs of the second DUT carrier unit 304 b
- an output ( 312 ) of the first DUT carrier unit 304 a is connected to a first input of second scan register 314 c
- an output of a second DUT carrier unit 304 b is connected to a first input of second scan registers 314 d, and so on.
- FIG. 4 illustrates some embodiments of a yield learning circuit structure 400 , comprising a defect under test (DUT) carrier unit 402 which comprises a plurality of logic devices 404 arranged in a logic tree structure, wherein a respective logic device 404 has been yield-hardened through heuristic yield analysis and physical failure analysis (PFA) to identify and correct defects within the DUT carrier unit 402 .
- the DUT carrier unit 402 comprises a plurality of DUT structures 406 , a respective DUT structure 406 comprising at least one process sensitive component which comprises a connection between respective logic devices 404 within the logic tree structure.
- the logic tree structure further comprises a final DUT structure 410 which outputs a measured value 414 of the yield learning circuit structure 400 .
- the final DUT structure 410 receives an input from an interior logic tree 412 , wherein a respective logic device 404 receives one or more inputs from a plurality of DUT structures 406 (i.e., respective logic device 404 has two inputs, one each from a respective DUT structure 406 , but in general may have three or more inputs).
- the interior logic tree 412 receives an input test vector (not shown) from first a shift register 416 comprising a plurality of scan registers 408 which are sequentially coupled and configured to receive the test vector as a serial data string and distribute the test vector in parallel format to the DUT carrier unit 402 .
- the serial data string comprises a multi-bit binary vector.
- a respective logic device 404 within the interior logic tree performs a logic operation (e.g., AND, OR, NAND, XOR, etc.) on one or more bits of the multi-bit binary vector.
- the yield learning circuit structure 400 sends the measured output 414 from the DUT carrier unit 402 as a single output bit through a second shift register 418 comprising a plurality of scan registers 408 which are sequentially coupled to a measurement and analysis tool.
- the second shift register 418 may also receive multiple output bits corresponding to multiple input test vectors, and distribute the multiple output bits as a serial data string to a measurement and analysis tool.
- FIG. 5 illustrates some embodiments of a DUT carrier circuit 500 , comprising logic devices which comprise a first XOR gate 502 , a second XOR gate 504 , and a third XOR gate 506 .
- a first output signal X 1 520 of a first scan register 508 is carried along a first connection 526 , which is coupled to a first input 532 of the first XOR gate 502 and a first input 534 of the second XOR gate 504 .
- a second output signal X 2 522 of a second scan register 510 is carried along a second connection 528 , which is coupled to a second input 540 of the second XOR gate 504 and a first input 536 of the third XOR gate 506 .
- a third output signal X 3 of a third scan register 512 is carried along a third connection 530 , which is coupled to a second input 538 of the first XOR gate 502 and a second input 542 of the third XOR gate 506 .
- a fourth output signal Q 1 550 of the first XOR gate 502 is carried along a fourth conducting path 544 which is coupled to a fourth scan register 514 .
- a fifth output signal Q 2 552 of the second XOR gate 504 carried along a fifth conducting path 546 which is coupled to a fifth scan register 516 .
- a sixth output signal Q 3 554 of the third XOR gate 506 is carried along a sixth conducting path 548 which is coupled to a sixth scan register 518 .
- a netlist comprising a definition of all connections between devices within the DUT carrier circuit 500 may be defined.
- the connections may be defined generally for any set of N DUT carrier circuit 500 by an N ⁇ N connectivity matrix which defines the inputs of the respective XOR gates of the DUT carrier circuit 500 , which operates on said inputs to produce expected outputs:
- a defect 556 is introduced in the DUT carrier circuit 500 which causes the value of X 1 to become “stuck” at 1.
- the first shift register 608 comprises a plurality of scan registers 610 which are sequentially coupled and configured to receive the test vector as a serial data string and distribute the test vector in parallel format to the to the 8-to-1 XOR tree 602 .
- a second shift register 612 comprising a plurality of scan registers 610 is configured to receive a measured output from DUT 604 O as a single measured output bit (not shown) for a respective input 8-bit binary vector.
- the 8-to-1 XOR tree 602 of FIG. 6A will produce an expected output for a given input test vector.
- a measured output that differs from the expected output is therefore due to a defect 614 (e.g., an open or short) in one or more of the DUT structures 604 A- 604 O within the DC yield learning circuit structure 600 A.
- Possible measured outputs may be simulated for a given input for a variety of defects formed within the DC yield learning circuit structure 600 A, and their results stored in a look-up table for use in identifying a location of the defect 614 .
- FIG. 6B illustrates some embodiments of a look-up table 600 B comprising expected outputs 624 vs. simulated results 626 for various input test vectors 616 A- 616 J from a collection of test vectors 618 to the DC yield learning circuit structure 600 A.
- an expected output 624 that differs from a corresponding simulated result 626 may be observed as a stuck-at 1 bit which comprises a value of 1 for any test vector 616 A- 616 J, or a stuck-at 0 bit which comprise a value of 0 for any test vector 616 A- 616 J.
- Stuck-at 0 bits corresponding to test vectors 616 A- 616 J may be observed in stuck-at 0 defect candidates 622 comprising a subset of the DUT structures 604 A- 604 O.
- Stuck-at 1 defect candidates 620 also comprise a subset of the DUT structures 604 A- 604 O (i.e., marked as “others”), and comprise the remaining subset of DUT structures 604 A- 604 O other than those shown in column 622 for a respective test vector 616 A- 616 J.
- an exact location (e.g., a failing DUT) may be identified by applying a plurality of input test vectors 616 A- 616 J wherein the expected output 624 differs from the simulated result 626 until a single common failing DUT is identified.
- Test vector 616 B produces an expected output 624 of 1, but a simulated result 626 of 0 for stuck-at 0 DUTs 604 B, 604 I, 604 M, and 604 O, or stuck at 1 DUTs 604 A, 604 C, 604 D, 604 E, 604 F, 604 G, 604 H, 604 J, 604 K, and 604 N.
- Test vector 616 I produces an expected output 624 of 0, but a simulated result 626 of 1 for stuck-at 0 DUTs 604 B, 604 D, 604 F, 604 H, 604 I, 604 J, 604 K, and 604 L, or stuck-at 1 DUTs 604 A, 604 C, 604 E, 604 G, 604 M, 604 N, and 604 O.
- Test vector 616 J produces an expected output 624 of 0, but a simulated result 626 of 1 for stuck-at 0 DUT structures 604 A, 604 C, 604 E, 604 G, 604 I, 604 J, 604 K, and 604 L, or stuck-at 1 DUTs 604 B, 604 D, 604 F, 604 H, 604 M, 604 N, and 604 O. Comparison of these four results for a common stuck-at 0 622 DUT results in DUT 604 I, and for a common stuck-at 1 620 DUT results in no DUT (because all 626 results are considered). Therefore, DUT 604 I has been identified containing the defect 614 .
- FIG. 7A illustrates some embodiments of an AC yield learning circuit structure 700 A, comprising a plurality of XOR gates 718 which have been yield-hardened though heuristic yield analysis.
- the plurality of XOR gates 718 comprise a logic tree structure with connections formed by a plurality of DUTs 720 .
- AC testing comprising transition testing may be performed at speed to detect transition violations resulting from crosstalk in the AC yield learning circuit structure 700 A resulting from node nets which are coupled. Crosstalk increases signal delay and can cause transition violations that can't be detected at DC frequencies (i.e., these cannot be detected in a stuck-at fault test).
- AC testing can also be used to detect faults resulting from defects which impact rise and fall times, set-up and hold times, and delay and access times at product application speeds.
- a respective pattern 802 - 814 of the discrete library of allowable patterns 800 A comprises a process sensitive component comprising a lithographic sensitive pattern formed on a via interconnect layer, one or more back end of line metallization layers, or a combination thereof.
- a lithographic sensitive pattern comprises a pattern of features formed on a layer which may or may not be resolvable by one or more lithographic tools which manufactures the layer.
- Examples of lithographic sensitive patterns include but are not limited to: two-dimensional features on a technology-minimum pitch, subminimum feature sizes, or multiple vias formed at or below technology-minimum space. These patterns may be manufactured and qualified through heuristic yield analysis.
- one or more parameters associated with a layout characteristic of a respective pattern may be varied systematically to gauge an impact on yield (e.g., width, space, etc.) and establish a lower-bound on the parameter for manufacturability.
- the one-layer comb structure 802 is formed on a first metallization layer 816 comprising a spine width W 1 , and comb width W 2 , and a comb space S 1 .
- the spine width W 1 and comb width W 2 may be systematically varied to create an open within the one-layer comb structure 802 .
- the comb space S 1 may be systematically varied to create to create a short to ground.
- the one-layer spacing structure 804 is formed on a first metallization layer 816 comprising a strip width W 3 which may be systematically varied to create an open, and a metal space S 2 which may be systematically varied to create to create a short to ground.
- the one-layer serpentine structure 806 is formed on a first metallization layer 816 comprising a serpentine width W 4 and serpentine space S 3 which may be systematically varied to create an open.
- These one-layer patterns 802 - 806 may be utilized to gauge failure modes associated with a resolution of one or more lithographic tools or processes. Other failure modes associated with manufacturing multiple metallization layers for lithographic tools or processes, or other manufacturing components may also be gauged by utilizing multi-layer patterns 808 - 814 from the discrete library of allowable patterns 800 A.
- the multi-layer interconnect structure 808 is formed on a first metallization layer 816 , and a second metallization layer 818 , with a via 820 that forms a connection between the first metallization layer 816 and the second metallization layer 818 .
- a setback E 1 of the via 820 within the first metallization layer 816 may be systematically varied to create an open.
- the multi-layer bridge structure 810 is formed on a first metallization layer 816 , and a second metallization layer 818 , with a via 820 that forms a connection between the first metallization layer 816 and the second metallization layer 818 .
- a bridge width W 5 of a wrong-way segment of the second metallization layer 818 may be systematically varied to create an open.
- the multi-layer shorting structure 812 is formed on multiple segments of a first metallization layer 816 and a on multiple segments of second metallization layer 818 , with a plurality of vias 820 forming various connections between the first metallization layer 816 and the second metallization layer 818 .
- a finger width W 6 of horizontal segments formed on the second metallization layer 818 , a first finger space S 4 between horizontal segments formed on the second metallization layer 818 , and a second finger space may all be systematically varied to create an open or short within the multi-layer shorting structure 812 .
- the multi-layer serpentine structure 814 is formed on multiple segments of a first metallization layer 816 and a on multiple segments of second metallization layer 818 , with a plurality of vias 820 forming various connections between the first metallization layer 816 and the second metallization layer 818 .
- the multi-layer serpentine structure 814 comprises a first metallization layer serpentine width W 8 , a second metallization layer serpentine width W 9 , and a serpentine space S 6 which may be systematically varied to create an open. Additionally, a setback (not shown) of one or more vias 820 within the first metallization layer 816 or the second metallization layer 818 may also be systematically varied to create an open.
- FIG. 8B illustrates some embodiments of a 4-to-1 XOR tree 800 B comprising a plurality of XOR gates 816 A- 816 C and the discrete library of allowable patterns 800 A.
- a respective element of discrete library of allowable patterns 800 A forms a connection between an output of a first XOR gate and an input of a second XOR gate as demonstrated by a connection 822 formed by the multi-layer bridge structure 810 between an output 824 of XOR gate 816 A and an input 826 of XOR gate 816 C.
- a netlist of the 4-to-1 XOR tree 800 B may be derived, and multi-bit test vectors may be written (i.e., 4-bit test vectors) and simulated to establish a look-up table (LUT) of expected outputs.
- the LUT may then be utilized to compare a measured output to an expected output when applying a respective test pattern and systematically varying one or more of the elements of the discrete library of allowable patterns 800 A which form connections between the XOR gates 816 A- 816 C of the 4-to-1 XOR tree 800 B, allowing for to determination that a respective element has developed a defect.
- an n-to-1 XOR tree may be utilized to qualify each pattern.
- a diagnosis framework to shorten yield learning cycles of technology node manufacturing processes from high defect density stage to technology maturity.
- a plurality of defect under test (DUT) structures are designed to capture potential manufacturing issues associated with defect formation.
- a test structure is formed by arranging the DUT structures within a DUT carrier unit, which has been yield-hardened though heuristic yield analysis such that a defect density of the DUT carrier unit is essentially zero. Possible outcomes of an application of test patterns and various failure scenarios associated with defects formed within the DUT structures within the DUT carrier unit are simulated and stored in a look-up table (LUT).
- LUT look-up table
- the LUT may then be referenced to determine a location of a defect within the test structure without the need for iterative analysis to correctly select defect candidates for physical failure analysis (PFA).
- PFA physical failure analysis
- a reusable defect under test (DUT) carrier unit is designed comprising a plurality of logic devices.
- a plurality of DUT structures are designed, wherein a respective DUT structure comprises at least one process sensitive component.
- a test structure is designed by arranging the plurality of DUT structures within the DUT carrier unit such that a respective DUT structure forms a connection between an output of a first logic device and an input of a second logic device (e.g., a logic tree).
- An expected output is simulated for a defect formed within the test structure, and stored in a look-up table (LUT). Heuristic yield analysis is applied to the test structure to identify the defect within the test structure.
- the look-up table is then utilized to locate the defect without performing time-consuming physical yield analysis (PFA).
- the present disclosure relates to a tool arrangement configured to locate and diagnose defects in an integrated circuit.
- a test vector generator is configured to define a test vector corresponding to a test structure.
- a fault simulation tool is configured to receive the test vector from the test vector generator and to simulate an expected output of applying the test vector to the test structure which corresponds to a failure location within the test structure.
- a look-up table is configured to receive the expected output from the fault simulation tool and to store the expected output.
- a measurement and analysis tool is configured to receive a measured output from applying the test vector to the test structure, and to compare the measured output to the expected output to determine a failure location within the test structure without performing physical failure analysis (PFA).
- a failure mode associated may either be determined through PFA or referenced from an earlier result.
- the present disclosure relates to a yield learning circuit structure comprising a defect under test (DUT) carrier unit and DUT structures.
- the DUT carrier unit comprises a plurality of logic devices arranged in a logic tree structure, wherein a respective logic device has been yield-hardened through heuristic yield analysis and physical failure analysis (PFA) to identify and correct defects within the DUT carrier unit.
- a respective DUT structure comprises at least one process sensitive component which comprises a connection between an output of a first logic device and an input of a second logic device within the logic tree structure.
- One of more parameters associated with the process sensitive component may be varied systematically within the DUT structure to gauge an impact on yield and establish a lower-bound on the parameter for manufacturability.
Abstract
Description
- Early technology node manufacturability is limited by yield loss due to high defect densities. Defects can result in immediate yield loss as well as downstream reliability issues. Heuristic yield analysis comprising the application of test patterns to an integrated circuit allows for observation of defects. However, diagnosis of a failure mode associated with a respective defect is a challenge as the location of the defect is difficult to determine. Often physical failure analysis is required to locate a defect and identify a failure mode.
-
FIG. 1 illustrates a flow diagram of some embodiments of a method to locate and diagnose defects in an integrated circuit. -
FIG. 2 illustrates some embodiments of a tool arrangement configured to locate and diagnose defects in an integrated circuit. -
FIG. 3 illustrates some embodiments of a yield learning circuit structure. -
FIG. 4 illustrates some embodiments of a yield learning circuit structure. -
FIG. 5 illustrates some embodiments of a DUT carrier circuit. -
FIG. 6A illustrates some embodiments of a DC yield learning circuit structure. -
FIG. 6B illustrates some embodiments of a look-up table comprising expected vs. simulated results -
FIG. 7A illustrates some embodiments of an AC yield learning circuit structure. -
FIG. 7B illustrates some embodiments of an AC look-up table. -
FIG. 8A illustrates some embodiments of a discrete library of allowable patterns for technology qualification comprising a process sensitive component. -
FIG. 8B illustrates some embodiments of a 4-to-1 XOR tree comprising a discrete library of allowable patterns. - The description herein is made with reference to the drawings, wherein like reference numerals are generally utilized to refer to like elements throughout, and wherein the various structures are not necessarily drawn to scale. In the following description, for purposes of explanation, numerous specific details are set forth in order to facilitate understanding. It may be evident, however, to one of ordinary skill in the art, that one or more aspects described herein may be practiced with a lesser degree of these specific details. In other instances, known structures and devices are shown in block diagram form to facilitate understanding.
- Diagnosis of defects in early technology node test vehicles allows for improved yield by determining a failure mode (i.e., root cause) associated with a manufacturing defect, which may be addressed through a change in design, a change in a manufacturing process, or a combination thereof. Current diagnosis methods utilize test patterns to observe defects, and allow for one to draw inferences based on observed behavior which can narrow the search for a failure mode. Typically, physical failure analysis (PFA) is utilized to determine a location and subsequently a failure mode associated with a respective defect. Moreover, several iterations of test pattern application, result interpretation, and making inferences may be necessary to determine a region where PFA may be utilized to determine an exact location and failure mode of a defect. This can limit the overall efficiency in diagnosing all failure modes, which can increase the lead time for product development and mask reliability issues which may only be discovered once a product is in the field. As such, increased efficiency in defect diagnosis is desired.
- Accordingly, the present disclosure relates to a diagnosis framework to shorten yield learning cycles of technology node manufacturing processes from a high defect density stage to technology maturity. A plurality of defect under test (DUT) structures are designed to capture potential manufacturing issues associated with defect formation. A test structure is formed by arranging the DUT structures within a defect under test (DUT) carrier unit, which has been yield-hardened though heuristic yield analysis such that a defect density of the DUT carrier unit is essentially zero. Possible outcomes of an application of test patterns and various failure scenarios associated with defects formed within the DUT structures within the DUT carrier unit are simulated and stored in a look-up table (LUT). The LUT may then be referenced to determine a location of a defect within the test structure without the need for iterative analysis to correctly select defect candidates for physical failure analysis (PFA). This diagnosis framework results in a higher resolution and rate of defect detection, shorter yield-ramp times, and optimal test resource.
-
FIG. 1 illustrates a flow diagram of some embodiments of amethod 100 to locate and diagnose defects within an integrated circuit. Whilemethod 100 is illustrated and described below as a series of acts or events, it will be appreciated that the illustrated ordering of such acts or events are not to be interpreted in a limiting sense. For example, some acts may occur in different orders and/or concurrently with other acts or events apart from those illustrated and/or described herein. In addition, not all illustrated acts may be required to implement one or more aspects or embodiments of the description herein. Further, one or more of the acts depicted herein may be carried out in one or more separate acts and/or phases. - At 102 a reusable defect under test (DUT) carrier unit comprising a plurality of logic devices is designed. The logic devices within the DUT carrier unit have been subjected to heuristic yield analysis comprising manufacture and test of the logic devices to identify defects formed within a respective logic device, physical failure analysis (PFA) to locate the defects and assign a failure mode to a respective defect, and have been yield-hardened through re-design, re-manufacture, and re-test. The heuristic yield analysis may comprise several iterations of re-design, re-manufacture, and re-test to insure robust yield. The DUT carrier unit may be generically designed and reused for an arbitrary collection of DUTs.
- At 104 a plurality of DUT structures are designed. A respective DUT structure comprises at least one process sensitive component formed on one or more of a front end of line gate material layer, a middle end of line interconnect layer, or a back end of line metallization layer.
- At 106 a test structure if designed by arranging the plurality of DUT structures within the DUT carrier unit, wherein a respective DUT structure forms a connection between an output of a first logic device and an input of a second logic device. The arrangement of the plurality of DUT structures within the DUT carrier unit comprise a logic tree structure, wherein a respective logic device receives its inputs from a plurality of DUTs and sends an output to a single DUT. The test structure is connected to one or more shift registers comprising a plurality of scan registers which are sequentially coupled, and allow for a scan-in of test vectors during testing.
- At 108 a netlist comprising a definition of connections between the logic devices within the test structure formed by the plurality of DUT structures is compiled. The netlist may be compiled using an industry-standard extraction tool or an in-house netlist compiler.
- At 110 a plurality of test vectors are generated for the test structure. A respective test vector comprises a multi-bit binary test vector which serves as an input to the test structure, and may be generated through utilization of an automatic test pattern generation (ATPG) tool to mimic a manufacturing defect within the test structure.
- At 112 expected outputs of applying the plurality of test vectors to the test structure are simulated to define possible failure locations within the test structure. The simulations may be performed using an industry-standard a fault simulation tool or design for test (DFT) tool, and are stored in a look-up table (LUT).
- At 114 the test structure is manufactured, which comprises fabrication of a silicon-based early technology yield learning vehicle in the 20 nm technology node, the 14 nm technology node, or lower technology node.
- At 116 the plurality test vectors are applied to the test structure to generate measured outputs. Application of the plurality test vectors comprises multi-bit binary test vectors scanned-in through a first shift register, operated on by the logic devices within the test structure to generate the measured outputs, which are then scanned out through a second shift register.
- At 118 the measured outputs are compared to the expected outputs to determine a failure location within the test structure, which comprises locating the stuck bits within the LUT.
- At 120 physical failure analysis (PFA) may be performed to the test structure at the failure location as defined by the LUT to determine a failure mode (e.g., an open, a short, etc.). The failure mode may be cataloged for future reference. Note that PFA is not necessary to determine the location of the fail as that is determined by the look-up table. Moreover, PFA need not be performed if the failure mode has already been identified and cataloged.
-
FIG. 2 illustrates some embodiments of atool arrangement 200 configured to locate and diagnose defects in an integrated circuit. Atest vector generator 204 is configured to define atest vector 206 corresponding to atest structure 202. Afault simulation tool 208 is configured to receive thetest vector 206 from thetest vector generator 204 and to simulate an expectedoutput 210 of applying thetest vector 206 to thetest structure 202 which corresponds to a failure location within thetest structure 202. A look-up table (LUT) 212 is configured to receive the expectedoutput 210 from thefault simulation tool 208 and to store the expectedoutput 210 for reference. A measurement andanalysis tool 214 is configured to receive a measuredoutput 216 of applying thetest vector 206 to thetest structure 202, and to compare the measuredoutput 216 to the expectedoutput 210 to determine a failure location within thetest structure 202 without performing physical failure analysis (PFA). - The
test structure 202 comprises afirst shift register 222 which is controlled by aclock 218 and configured to receive thetest vector 206 from thetest vector generator 204. Acontroller 220 is configured to control operation of thetest vector generator 204 and theclock 218. Thefirst shift register 222 comprises a plurality of scan registers 226 which are sequentially coupled and configured to receive thetest vector 206 comprising a multi-bit binary vector as a serial data string and distribute thetest vector 206 in parallel format to aDUT carrier unit 224. Asecond shift register 228 which is controlled by theclock 218 is configured to receive a measuredoutput 216 from the DUT carrier unit, and to send the measuredoutput 216 to the measurement andanalysis tool 214. Thesecond shift register 228 comprises a plurality of scan registers 226 which are sequentially coupled and configured to receive the measuredoutput 216 from theDUT carrier unit 224 as a single bit, and to distribute the measuredoutput 216 to a measurement andanalysis tool 214. Thesecond shift register 228 may also receive multiple measuredoutputs 216 and distribute the multiple measuredoutputs 216 as a serial data string to a measurement andanalysis tool 214. - The
DUT carrier unit 224 comprises a plurality of logic devices which are coupled by a plurality of DUT structures to form a logic tree for which a known input (e.g., the test vector 206) will produce an expectedoutput 210. It will be appreciated in subsequent embodiments that through heuristic yield analysis and PFA to identify and correct defects that thefirst shift register 222, thesecond shift register 228, and the plurality of logic devices within thetest structure 202 may be yield-hardened such that defect formation is limited to the DUT structures. It will be further appreciated that a structure of the logic tree will allow for the expectedoutput 210 to be used to determine a location of a defect corresponding to DUT structure. - Before the
test structure 202 can be tested it must first be designed and manufactured. Adesign tool 230 is configured to produce a design of thetest structure 202, and is coupled to anetlist compiler 232 configured to receive information about the design of thetest structure 202 and to compile a netlist of the connections formed between the plurality of logic devices. Thenetlist compiler 232 is further configured to send the netlist to thefault simulation tool 208 in order to simulate the expectedoutput 210. - It will be appreciated that the structure of the logic tree within the
test structure 202 allows for the expectedoutput 210 to determine a location of a defect (e.g., a defective DUT) within thetest structure 202 when compared to the measuredoutput 216 in the measurement andanalysis tool 214. Knowledge of the location of the defect may be sufficient for many testing applications. However, it may also be useful to assign a failure mode (i.e., a root cause of the failure) by subjecting thetest structure 202 to destructive physical failure analysis (PFA) in aPFA tool 234. While in some prior art approaches PFA is utilized as a component to aid in determination the location of the defect, the measurement andanalysis tool 214 of thetool arrangement 200 can determine the location without PFA by utilizing the look-up table 212 to compare the expectedoutput 210 to the measuredoutput 216. -
FIG. 3 illustrates some embodiments of a yieldlearning circuit structure 300, comprising afirst shift register 301 a, asecond shift register 301 b, and a plurality of defect under test (DUT) carrier units 304 a-304 j, wherein a respective DUT carrier unit of the plurality of DUT carrier units 304 a-304 j comprises a plurality of logic devices arranged in a logic tree structure which has been yield-hardened. Thefirst shift register 301 a comprises a plurality of first scan registers 302 a-302 o which are sequentially coupled, wherein a respective first scan register of the plurality of first scan registers 302 a-302 o is connected to N DUT carrier units of the plurality of DUT carrier units 304 a-304 j. For the embodiments of a yield learning circuit structure 300 a respective DUT carrier unit of the plurality of DUT carrier units 304 a-304 j comprises an 8-to-1 XOR tree with 8 inputs and 1 output (i.e., N=8). Each output of a respective DUT carrier unit of the plurality of DUT carrier units 304 a-304 j is connected to a respective second scan register of a plurality of second scan registers 314 a-314 o of thesecond shift register 301 b, which selects an input source according tomode selection signal 310. - Each second scan register 314 a-314 o comprises two data inputs. A first input of a respective second scan register of the plurality of second scan registers 314 a-314 o is connected to an output of a respective DUT carrier unit of the plurality of DUT carrier units 304 a-304 j. A second input of a respective second scan register of the plurality of second scan registers 314 a-314 o is connected to an output of a previous second scan register of the plurality of second scan registers 314 a-314 o (i.e., a second input of scan register 314 o is connected to an output of
scan register 314 n, etc.). Thefirst shift register 301 a and thesecond shift register 301 b each comprise two operation modes: a shift mode and a capture mode, which are controlled by amode selection signal 310. As a non-limiting example, in shift mode (mode selection signal=1), all second scan registers of the plurality of second scan registers 314 a-314 o form a shift register between a primary input and a primary output, and each second scan register of the plurality of second scan registers 314 a-314 o will receive an output value of a preceding second scan register of the plurality of second scan registers 314 a-314 o. In capture mode (mode selection signal=0), each second scan register of the plurality of second scan registers 314 a-314 o will change to receive the output value of a respective DUT carrier unit of the plurality of DUT carrier units 304 a-304 j. - For a yield
learning circuit structure 300 comprising M DUT carrier units 304 a-304 j, thefirst shift register 301 a will require M first scan registers of the plurality of first scan registers 302 a-302 o, and thesecond shift register 301 b will each require M second scan registers of the plurality of second scan registers 314 a-314 o. As a non-limiting example, for an 8-input firstDUT carrier unit 304 a,output data 312 of the first 8 first scan registers 302 a-302 j are connected to 8 inputs of the firstDUT carrier unit 304 a, outputs of the 2nd through 9th first scan registers 302 b-302 k are connected to the 8 inputs of the secondDUT carrier unit 304 b, and so on. On the output side, an output (312) of the firstDUT carrier unit 304 a is connected to a first input ofsecond scan register 314 c, an output of a secondDUT carrier unit 304 b is connected to a first input of second scan registers 314 d, and so on. -
FIG. 4 illustrates some embodiments of a yieldlearning circuit structure 400, comprising a defect under test (DUT)carrier unit 402 which comprises a plurality oflogic devices 404 arranged in a logic tree structure, wherein arespective logic device 404 has been yield-hardened through heuristic yield analysis and physical failure analysis (PFA) to identify and correct defects within theDUT carrier unit 402. TheDUT carrier unit 402 comprises a plurality ofDUT structures 406, arespective DUT structure 406 comprising at least one process sensitive component which comprises a connection betweenrespective logic devices 404 within the logic tree structure. The logic tree structure further comprises afinal DUT structure 410 which outputs a measuredvalue 414 of the yieldlearning circuit structure 400. Thefinal DUT structure 410 receives an input from aninterior logic tree 412, wherein arespective logic device 404 receives one or more inputs from a plurality of DUT structures 406 (i.e.,respective logic device 404 has two inputs, one each from arespective DUT structure 406, but in general may have three or more inputs). Theinterior logic tree 412 receives an input test vector (not shown) from first ashift register 416 comprising a plurality of scan registers 408 which are sequentially coupled and configured to receive the test vector as a serial data string and distribute the test vector in parallel format to theDUT carrier unit 402. - The serial data string comprises a multi-bit binary vector. A
respective logic device 404 within the interior logic tree performs a logic operation (e.g., AND, OR, NAND, XOR, etc.) on one or more bits of the multi-bit binary vector. The yieldlearning circuit structure 400 sends the measuredoutput 414 from theDUT carrier unit 402 as a single output bit through asecond shift register 418 comprising a plurality of scan registers 408 which are sequentially coupled to a measurement and analysis tool. Thesecond shift register 418 may also receive multiple output bits corresponding to multiple input test vectors, and distribute the multiple output bits as a serial data string to a measurement and analysis tool. -
FIG. 5 illustrates some embodiments of aDUT carrier circuit 500, comprising logic devices which comprise afirst XOR gate 502, asecond XOR gate 504, and athird XOR gate 506. A firstoutput signal X1 520 of afirst scan register 508 is carried along afirst connection 526, which is coupled to afirst input 532 of thefirst XOR gate 502 and afirst input 534 of thesecond XOR gate 504. A secondoutput signal X2 522 of asecond scan register 510 is carried along asecond connection 528, which is coupled to asecond input 540 of thesecond XOR gate 504 and afirst input 536 of thethird XOR gate 506. A third output signal X3 of athird scan register 512 is carried along athird connection 530, which is coupled to asecond input 538 of thefirst XOR gate 502 and asecond input 542 of thethird XOR gate 506. A fourthoutput signal Q1 550 of thefirst XOR gate 502 is carried along afourth conducting path 544 which is coupled to afourth scan register 514. A fifthoutput signal Q2 552 of thesecond XOR gate 504 carried along afifth conducting path 546 which is coupled to afifth scan register 516. A sixthoutput signal Q3 554 of thethird XOR gate 506 is carried along asixth conducting path 548 which is coupled to asixth scan register 518. - To determine possible locations of defects that may form within a respective DUT coupled to the
DUT carrier circuit 500, a netlist comprising a definition of all connections between devices within theDUT carrier circuit 500 may be defined. The connections, as described in the previous paragraph, may defined generally for any set of NDUT carrier circuit 500 by an N×N connectivity matrix which defines the inputs of the respective XOR gates of theDUT carrier circuit 500, which operates on said inputs to produce expected outputs: -
- For the
DUT carrier circuit 500 there are N=3 connections. Therefore, the above operation becomes: -
- which gives:
- Therefore, for the inputs X1, X2, and X3 comprising values of 0 or 1, the outputs Q1, Q2, and Q3 comprising values of 0 or 1 are fully-determined for the
DUT carrier circuit 500 by the connectivity matrix and function of the logic devices. This allows for simulation of expected outputs when applying the inputs X1, X2, and X3. - In this exemplary embodiment of the
DUT carrier circuit 500, adefect 556 is introduced in theDUT carrier circuit 500 which causes the value of X1 to become “stuck” at 1. As a result, an input test vector of (X1,X2,X3)=(0,1,1) would give an expected output of: -
- but because X1 is stuck at 1, a measured output is:
-
- The measured output (Q1,Q2,Q3)=(0,0,0) can be anticipated by modifying the input test vector to (X1,X2,X3)=(1,1,1).
- Therefore, applying a test vector to the
DUT carrier circuit 500 to generate the measured output, and comparing the expected output to the measured output allows for identification of a defect. Since the connectivity matrix (e.g., netlist) and the function of the logic devices comprising theDUT carrier circuit 500 are known, the result of any combination of values for the input test vector (X1,X2,X3) can be simulated and stored (i.e., similar to a look-up table), thus allowing one to determine a respective failure-location within theDUT carrier circuit 500 without the need for destructive failure analysis. Moreover, thefirst XOR gate 502,second XOR gate 504, andthird XOR gate 506 of theDUT carrier circuit 500 comprise first through third DUT carrier units respectively, wherein a respective carrier unit comprises a 2-to-1 XOR tree. Note that the size of theDUT carrier circuit 500 is scalable and extendable. For example, an 8×8 connectivity matrix to describe connectivity of a DUT carrier unit comprising an 8-to-1 XOR tree may be derived by a similar analysis. The embodiments of aDUT carrier circuit 500 can be used to understand the basic diagnosis concepts and matrix construction of a netlist. In general this analysis can be extended to a DUT carrier circuit with an arbitrary number of connections. -
FIG. 6A illustrates some embodiments of a DC yieldlearning circuit structure 600A, comprising an 8-to-1XOR tree 602 configured to carry 15DUT structures 604A-604O. A respective DUT structure from the 15DUT structures 604A-604O of the 8-to-1XOR tree 602 comprises a connection between an output of afirst XOR gate 606 and an input of asecond XOR gate 606, wherein eachXOR gate 606 is coupled to two DUT structures from the 15DUT structures 604A-604O which form input connections to theXOR gate 606. Afirst shift register 608 delivers a test vector (not shown) comprising an 8-bit binary vector to the 8-to-1XOR tree 602. Thefirst shift register 608 comprises a plurality of scan registers 610 which are sequentially coupled and configured to receive the test vector as a serial data string and distribute the test vector in parallel format to the to the 8-to-1XOR tree 602. Asecond shift register 612 comprising a plurality of scan registers 610 is configured to receive a measured output from DUT 604O as a single measured output bit (not shown) for a respective input 8-bit binary vector. - The 8-to-1
XOR tree 602 ofFIG. 6A will produce an expected output for a given input test vector. A measured output that differs from the expected output is therefore due to a defect 614 (e.g., an open or short) in one or more of theDUT structures 604A-604O within the DC yieldlearning circuit structure 600A. Possible measured outputs may be simulated for a given input for a variety of defects formed within the DC yieldlearning circuit structure 600A, and their results stored in a look-up table for use in identifying a location of thedefect 614. -
FIG. 6B illustrates some embodiments of a look-up table 600B comprising expectedoutputs 624 vs.simulated results 626 for variousinput test vectors 616A-616J from a collection oftest vectors 618 to the DC yieldlearning circuit structure 600A. For arespective test vector 616A-616J, an expectedoutput 624 that differs from a correspondingsimulated result 626 may be observed as a stuck-at 1 bit which comprises a value of 1 for anytest vector 616A-616J, or a stuck-at 0 bit which comprise a value of 0 for anytest vector 616A-616J. Stuck-at 0 bits corresponding to testvectors 616A-616J may be observed in stuck-at 0defect candidates 622 comprising a subset of theDUT structures 604A-604O. Stuck-at 1defect candidates 620 also comprise a subset of theDUT structures 604A-604O (i.e., marked as “others”), and comprise the remaining subset ofDUT structures 604A-604O other than those shown incolumn 622 for arespective test vector 616A-616J. For example, fortest vector 616A the “others” (620) comprise 604 x, where x=B,C,D,E,F,G,H,J,K,L,N. For either stuck-at 1defect candidates 620 or stuck-at 0defect candidates 622, an exact location (e.g., a failing DUT) may be identified by applying a plurality ofinput test vectors 616A-616J wherein the expectedoutput 624 differs from thesimulated result 626 until a single common failing DUT is identified. Note that for the embodiments of the look-up table 600B a respective stuck-at 0defect candidate 622 comprising a letter x=A-O which corresponds to a givenDUT structure 604 x ofFIG. 6A . - For the embodiments of the look-up table 600B corresponding to the DC yield
learning circuit structure 600A, wherein DUT 604I has been compromised due to thedefect 614, thetest vector 616A produces an expectedoutput 624 of 1, but asimulated result 626 of 0. Therefore, one or more ofDUT structures DUT structures Test vector 616B produces an expectedoutput 624 of 1, but asimulated result 626 of 0 for stuck-at 0DUTs DUTs output 624 of 0, but asimulated result 626 of 1 for stuck-at 0DUTs DUTs Test vector 616J produces an expectedoutput 624 of 0, but asimulated result 626 of 1 for stuck-at 0DUT structures DUTs defect 614. - The DC yield
learning circuit structure 600A and corresponding look-up table 600B comprise some embodiments of a device and method to identify and locate DC fails within the DC yieldlearning circuit structure 600A. Stuck-at 0 bits comprise a structure that has become shorted to ground or an open has formed due to a manufacturing defect. Stuck-at 1 bits comprise a structure that has become shorted to VDD due to a manufacturing defect. Stuck-at 0 bits and stuck-at 1 bits comprise stuck-at fault bits which can be identified in DC testing. Failure modes identified during DC testing comprise hard fails that are not related to timing (e.g., frequency) and can reduce yield. However, soft fail mechanisms may only be identified during AC testing, comprising defects related to timing. Coupling faults, wherein changing a state of a first DUT changes the state of a second DUT, and transition faults, wherein a DUT fails to undergo a 0 1 or a 1 0 transition are two examples of soft fails which may only be identified during AC testing. Pattern sensitive faults are of particular interest in lower technology nodes due to limitations in current lithographic tools in resolving two-dimensional features on a technology-minimum pitch. -
FIG. 7A illustrates some embodiments of an AC yieldlearning circuit structure 700A, comprising a plurality ofXOR gates 718 which have been yield-hardened though heuristic yield analysis. The plurality ofXOR gates 718 comprise a logic tree structure with connections formed by a plurality ofDUTs 720. AC testing comprising transition testing may be performed at speed to detect transition violations resulting from crosstalk in the AC yieldlearning circuit structure 700A resulting from node nets which are coupled. Crosstalk increases signal delay and can cause transition violations that can't be detected at DC frequencies (i.e., these cannot be detected in a stuck-at fault test). AC testing can also be used to detect faults resulting from defects which impact rise and fall times, set-up and hold times, and delay and access times at product application speeds. -
FIG. 7B illustrates some embodiments of an AC look-up table 700B comprising an index of failingAC patterns 724B which may be utilized to locate an AC fault observed within a respective pin group 702-714, oroutput pin 716 ofFIG. 7A , resulting from a measured output that differs from an expected output. -
FIG. 8A illustrates some embodiments of a discrete library ofallowable patterns 800A for technology qualification comprising seven elements: a one-layer comb structure 802, a one-layer spacing structure 804, a one-layer serpentine structure 806, amulti-layer interconnect structure 808, amulti-layer bridge structure 810, amulti-layer shorting structure 812, and a multi-layerserpentine structure 814. It will be appreciated that these exemplary embodiments of the discrete library ofallowable patterns 800A comprise a non-limiting example for the purpose of understanding, and that additional patterns may also be included into the discrete library ofallowable patterns 800A. - A respective pattern 802-814 of the discrete library of
allowable patterns 800A comprises a process sensitive component comprising a lithographic sensitive pattern formed on a via interconnect layer, one or more back end of line metallization layers, or a combination thereof. It will be appreciated that a lithographic sensitive pattern comprises a pattern of features formed on a layer which may or may not be resolvable by one or more lithographic tools which manufactures the layer. Examples of lithographic sensitive patterns include but are not limited to: two-dimensional features on a technology-minimum pitch, subminimum feature sizes, or multiple vias formed at or below technology-minimum space. These patterns may be manufactured and qualified through heuristic yield analysis. Moreover, one or more parameters associated with a layout characteristic of a respective pattern may be varied systematically to gauge an impact on yield (e.g., width, space, etc.) and establish a lower-bound on the parameter for manufacturability. - The one-
layer comb structure 802 is formed on afirst metallization layer 816 comprising a spine width W1, and comb width W2, and a comb space S1. The spine width W1 and comb width W2 may be systematically varied to create an open within the one-layer comb structure 802. The comb space S1 may be systematically varied to create to create a short to ground. The one-layer spacing structure 804 is formed on afirst metallization layer 816 comprising a strip width W3 which may be systematically varied to create an open, and a metal space S2 which may be systematically varied to create to create a short to ground. The one-layer serpentine structure 806 is formed on afirst metallization layer 816 comprising a serpentine width W4 and serpentine space S3 which may be systematically varied to create an open. These one-layer patterns 802-806 may be utilized to gauge failure modes associated with a resolution of one or more lithographic tools or processes. Other failure modes associated with manufacturing multiple metallization layers for lithographic tools or processes, or other manufacturing components may also be gauged by utilizing multi-layer patterns 808-814 from the discrete library ofallowable patterns 800A. - The
multi-layer interconnect structure 808 is formed on afirst metallization layer 816, and asecond metallization layer 818, with a via 820 that forms a connection between thefirst metallization layer 816 and thesecond metallization layer 818. A setback E1 of the via 820 within thefirst metallization layer 816 may be systematically varied to create an open. Themulti-layer bridge structure 810 is formed on afirst metallization layer 816, and asecond metallization layer 818, with a via 820 that forms a connection between thefirst metallization layer 816 and thesecond metallization layer 818. A bridge width W5 of a wrong-way segment of thesecond metallization layer 818 may be systematically varied to create an open. Themulti-layer shorting structure 812 is formed on multiple segments of afirst metallization layer 816 and a on multiple segments ofsecond metallization layer 818, with a plurality ofvias 820 forming various connections between thefirst metallization layer 816 and thesecond metallization layer 818. A finger width W6 of horizontal segments formed on thesecond metallization layer 818, a first finger space S4 between horizontal segments formed on thesecond metallization layer 818, and a second finger space may all be systematically varied to create an open or short within themulti-layer shorting structure 812. The multi-layerserpentine structure 814 is formed on multiple segments of afirst metallization layer 816 and a on multiple segments ofsecond metallization layer 818, with a plurality ofvias 820 forming various connections between thefirst metallization layer 816 and thesecond metallization layer 818. The multi-layerserpentine structure 814 comprises a first metallization layer serpentine width W8, a second metallization layer serpentine width W9, and a serpentine space S6 which may be systematically varied to create an open. Additionally, a setback (not shown) of one ormore vias 820 within thefirst metallization layer 816 or thesecond metallization layer 818 may also be systematically varied to create an open. -
FIG. 8B illustrates some embodiments of a 4-to-1XOR tree 800B comprising a plurality ofXOR gates 816A-816C and the discrete library ofallowable patterns 800A. A respective element of discrete library ofallowable patterns 800A forms a connection between an output of a first XOR gate and an input of a second XOR gate as demonstrated by aconnection 822 formed by themulti-layer bridge structure 810 between anoutput 824 ofXOR gate 816A and aninput 826 ofXOR gate 816C. A netlist of the 4-to-1XOR tree 800B may be derived, and multi-bit test vectors may be written (i.e., 4-bit test vectors) and simulated to establish a look-up table (LUT) of expected outputs. The LUT may then be utilized to compare a measured output to an expected output when applying a respective test pattern and systematically varying one or more of the elements of the discrete library ofallowable patterns 800A which form connections between theXOR gates 816A-816C of the 4-to-1XOR tree 800B, allowing for to determination that a respective element has developed a defect. Note that in general for a discrete library of 2n-1 of allowable patterns, an n-to-1 XOR tree may be utilized to qualify each pattern. - It will also be appreciated that equivalent alterations and/or modifications may occur to one of ordinary skill in the art based upon a reading and/or understanding of the specification and annexed drawings. The disclosure herein includes all such modifications and alterations and is generally not intended to be limited thereby. In addition, while a particular feature or aspect may have been disclosed with respect to only one of several implementations, such feature or aspect may be combined with one or more other features and/or aspects of other implementations as may be desired. Furthermore, to the extent that the terms “includes”, “having”, “has”, “with”, and/or variants thereof are used herein; such terms are intended to be inclusive in meaning—like “comprising.” Also, “exemplary” is merely meant to mean an example, rather than the best. It is also to be appreciated that features, layers and/or elements depicted herein are illustrated with particular dimensions and/or orientations relative to one another for purposes of simplicity and ease of understanding, and that the actual dimensions and/or orientations may differ substantially from that illustrated herein.
- Therefore, the present disclosure relates to a diagnosis framework to shorten yield learning cycles of technology node manufacturing processes from high defect density stage to technology maturity. A plurality of defect under test (DUT) structures are designed to capture potential manufacturing issues associated with defect formation. A test structure is formed by arranging the DUT structures within a DUT carrier unit, which has been yield-hardened though heuristic yield analysis such that a defect density of the DUT carrier unit is essentially zero. Possible outcomes of an application of test patterns and various failure scenarios associated with defects formed within the DUT structures within the DUT carrier unit are simulated and stored in a look-up table (LUT). The LUT may then be referenced to determine a location of a defect within the test structure without the need for iterative analysis to correctly select defect candidates for physical failure analysis (PFA). This diagnosis framework results in a higher resolution and rate of defect detection, shorter yield-ramp times, and optimal test resource.
- In some embodiments the present disclosure relates to a method to locate and diagnose defects in an integrated circuit. A reusable defect under test (DUT) carrier unit is designed comprising a plurality of logic devices. A plurality of DUT structures are designed, wherein a respective DUT structure comprises at least one process sensitive component. A test structure is designed by arranging the plurality of DUT structures within the DUT carrier unit such that a respective DUT structure forms a connection between an output of a first logic device and an input of a second logic device (e.g., a logic tree). An expected output is simulated for a defect formed within the test structure, and stored in a look-up table (LUT). Heuristic yield analysis is applied to the test structure to identify the defect within the test structure. The look-up table is then utilized to locate the defect without performing time-consuming physical yield analysis (PFA).
- In some embodiments the present disclosure relates to a tool arrangement configured to locate and diagnose defects in an integrated circuit. A test vector generator is configured to define a test vector corresponding to a test structure. A fault simulation tool is configured to receive the test vector from the test vector generator and to simulate an expected output of applying the test vector to the test structure which corresponds to a failure location within the test structure. A look-up table is configured to receive the expected output from the fault simulation tool and to store the expected output. A measurement and analysis tool is configured to receive a measured output from applying the test vector to the test structure, and to compare the measured output to the expected output to determine a failure location within the test structure without performing physical failure analysis (PFA). A failure mode associated may either be determined through PFA or referenced from an earlier result.
- In some embodiments the present disclosure relates to a yield learning circuit structure comprising a defect under test (DUT) carrier unit and DUT structures. The DUT carrier unit comprises a plurality of logic devices arranged in a logic tree structure, wherein a respective logic device has been yield-hardened through heuristic yield analysis and physical failure analysis (PFA) to identify and correct defects within the DUT carrier unit. A respective DUT structure comprises at least one process sensitive component which comprises a connection between an output of a first logic device and an input of a second logic device within the logic tree structure. One of more parameters associated with the process sensitive component may be varied systematically within the DUT structure to gauge an impact on yield and establish a lower-bound on the parameter for manufacturability.
Claims (20)
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/588,155 US9310431B2 (en) | 2012-08-17 | 2012-08-17 | Diagnosis framework to shorten yield learning cycles of advanced processes |
SG2012066833A SG2012066833A (en) | 2012-08-17 | 2012-09-10 | A diagnosis framework to shorten yield learning cycles of advanced processes |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/588,155 US9310431B2 (en) | 2012-08-17 | 2012-08-17 | Diagnosis framework to shorten yield learning cycles of advanced processes |
Publications (2)
Publication Number | Publication Date |
---|---|
US20140049281A1 true US20140049281A1 (en) | 2014-02-20 |
US9310431B2 US9310431B2 (en) | 2016-04-12 |
Family
ID=50099630
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US13/588,155 Active 2035-01-28 US9310431B2 (en) | 2012-08-17 | 2012-08-17 | Diagnosis framework to shorten yield learning cycles of advanced processes |
Country Status (2)
Country | Link |
---|---|
US (1) | US9310431B2 (en) |
SG (1) | SG2012066833A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190259698A1 (en) * | 2018-02-17 | 2019-08-22 | Sandisk Technologies Llc | Interconnects containing serpentine line structures for three-dimensional memory devices and methods of making the same |
CN111896865A (en) * | 2020-07-30 | 2020-11-06 | 电子科技大学 | Fault position detection method for signal acquisition system |
CN114155367A (en) * | 2022-02-09 | 2022-03-08 | 北京阿丘科技有限公司 | Method, device and equipment for detecting defects of printed circuit board and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6744268B2 (en) * | 1998-08-27 | 2004-06-01 | The Micromanipulator Company, Inc. | High resolution analytical probe station |
US20080099762A1 (en) * | 2006-11-01 | 2008-05-01 | Synopsys, Inc. | Differential voltage defectivity monitoring circuit |
US20130285739A1 (en) * | 2010-09-07 | 2013-10-31 | Corporation De L ' Ecole Polytechnique De Montreal | Methods, apparatus and system to support large-scale micro- systems including embedded and distributed power supply, thermal regulation, multi-distributedsensors and electrical signal propagation |
-
2012
- 2012-08-17 US US13/588,155 patent/US9310431B2/en active Active
- 2012-09-10 SG SG2012066833A patent/SG2012066833A/en unknown
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6744268B2 (en) * | 1998-08-27 | 2004-06-01 | The Micromanipulator Company, Inc. | High resolution analytical probe station |
US20080099762A1 (en) * | 2006-11-01 | 2008-05-01 | Synopsys, Inc. | Differential voltage defectivity monitoring circuit |
US20130285739A1 (en) * | 2010-09-07 | 2013-10-31 | Corporation De L ' Ecole Polytechnique De Montreal | Methods, apparatus and system to support large-scale micro- systems including embedded and distributed power supply, thermal regulation, multi-distributedsensors and electrical signal propagation |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190259698A1 (en) * | 2018-02-17 | 2019-08-22 | Sandisk Technologies Llc | Interconnects containing serpentine line structures for three-dimensional memory devices and methods of making the same |
US10553537B2 (en) * | 2018-02-17 | 2020-02-04 | Sandisk Technologies Llc | Interconnects containing serpentine line structures for three-dimensional memory devices and methods of making the same |
CN111896865A (en) * | 2020-07-30 | 2020-11-06 | 电子科技大学 | Fault position detection method for signal acquisition system |
CN114155367A (en) * | 2022-02-09 | 2022-03-08 | 北京阿丘科技有限公司 | Method, device and equipment for detecting defects of printed circuit board and storage medium |
Also Published As
Publication number | Publication date |
---|---|
SG2012066833A (en) | 2014-03-28 |
US9310431B2 (en) | 2016-04-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20190311290A1 (en) | Deep Learning Based Test Compression Analyzer | |
CN105631077B (en) | Integrated circuit with increased fault coverage | |
US7870519B2 (en) | Method for determining features associated with fails of integrated circuits | |
US11361248B2 (en) | Multi-stage machine learning-based chain diagnosis | |
US11041906B2 (en) | Optimized scan chain diagnostic pattern generation for reversible scan architecture | |
US10657207B1 (en) | Inter-cell bridge defect diagnosis | |
US11156661B2 (en) | Reversible multi-bit scan cell-based scan chains for improving chain diagnostic resolution | |
US11585853B2 (en) | Trajectory-optimized test pattern generation for built-in self-test | |
US10509072B2 (en) | Test application time reduction using capture-per-cycle test points | |
US11073556B2 (en) | Low pin count reversible scan architecture | |
EP3756020B1 (en) | Deterministic stellar built-in self-test | |
US7219287B1 (en) | Automated fault diagnosis in a programmable device | |
US9310431B2 (en) | Diagnosis framework to shorten yield learning cycles of advanced processes | |
US10830815B2 (en) | Signal probability-based test cube reordering and merging | |
US10078720B2 (en) | Methods and systems for circuit fault diagnosis | |
US11681843B2 (en) | Input data compression for machine learning-based chain diagnosis | |
US11408938B2 (en) | Bidirectional scan cells for single-path reversible scan chains | |
d'Hondt et al. | A comprehensive framework for cell-aware diagnosis of customer returns | |
Jahangiri et al. | Value-added defect testing techniques | |
Cook et al. | Diagnosis of multiple faults with highly compacted test responses | |
Wunderlich | From embedded test to embedded diagnosis | |
US11092645B2 (en) | Chain testing and diagnosis using two-dimensional scan architecture | |
US7188043B1 (en) | Boundary scan analysis | |
Khatri et al. | Development and Verification of Serial Fault Simulation for FPGA Designs using the Proposed RASP-FIT Tool | |
Girard et al. | Defect Diagnosis Techniques for Silicon Customer Returns |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: TAIWAN SEMICONDUCTOR MANUFACTURING CO., LTD., TAIW Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:LIU, YEN-LING;TSENG, NAN-HSIN;CHEN, JI-JAN;AND OTHERS;REEL/FRAME:028804/0289 Effective date: 20120816 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
MAFP | Maintenance fee payment |
Free format text: PAYMENT OF MAINTENANCE FEE, 4TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1551); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY Year of fee payment: 4 |
|
FEPP | Fee payment procedure |
Free format text: MAINTENANCE FEE REMINDER MAILED (ORIGINAL EVENT CODE: REM.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |